Skip to main content
4 events
when toggle format what by license comment
Jan 1, 2014 at 11:01 comment added usr Maybe map does some clever allocation strategy. If every tree node was a malloc-backed allocation there would be lots of overhead. Good to know.
Jan 1, 2014 at 4:55 comment added microtherion Interesting thought. In my tests (clang on a Mac, latest C++ library), std::map<uint32_t,float> consistently uses less memory than std::unordered_map<uint32_t,float> (presumably because hash tables need a certain amount of free space to give decent performance). On the other hand, std::unordered_map was about twice as fast.
Jan 1, 2014 at 0:07 comment added usr A tree-based map likely has more memory usage than a hashtable based container. I'd use a hashtable if available.
Jun 5, 2013 at 23:51 history answered microtherion CC BY-SA 3.0